摘要
讨论了一类针对传统LMS算法进行改进的变步长自适应算法,分析其性能,对原有算法进行改进,并针对输入信号高度相关时算法收敛速度下降导致性能下降的问题,引入了解相关原理,用输入向量的正交分量来更新滤波器权系数,有效加快了算法的收敛速度,并保持了原算法的良好性能。
A series of variable step - size adaptive algorithms are proposed to improve performance of conventional least mean square （LMS） algorithm. In order to solve performance reduction due to lower convergence speed of the algorithms, decorrelation principle is introduced, that is, to update＇weight coefficient of filter by using of quadrature component of input vector, which could effectively increase convergence speed of the algorithms, meanwhile good performance of the original algorithm is maintained.
出处
《火控雷达技术》
2009年第3期47-50,共4页
Fire Control Radar Technology
关键词
自适应算法
变步长
解相关
LMS
adaptive algorithm
variable step-size
decorrelation
least mean square （LMS）
